Sparse Uncertainty Representation in Deep Learning with Inducing Weights

Hippolyt Ritter, Martin Kukla, Cheng Zhang, Yingzhen Li

Keywords: Bayesian neural networks, uncertainty estimation

TL;DR: For the first time, reducing parameter count of BNNs & deep ensembles to be < 1/4 of a deterministic network.

Abstract: Bayesian Neural Networks and deep ensembles represent two modern paradigms of uncertainty quantification in deep learning. Yet these approaches struggle to scale mainly due to memory inefficiency, requiring parameter storage several times that of their deterministic counterparts. To address this, we augment each weight matrix with a small inducing weight matrix, projecting the uncertainty quantification into a lower dimensional space. We further extend Matheron’s conditional Gaussian sampling rule to enable fast weight sampling, which enables our inference method to maintain reasonable run-time as compared with ensembles. Importantly, our approach achieves competitive performance to the state-of-the-art in prediction and uncertainty estimation tasks with fully connected neural networks and ResNets, while reducing the parameter size to $\leq 24.3\%$ of that of a single neural network.

Code Of Conduct: I certify that all co-authors of this work have read and commit to adhering to the NeurIPS Statement on Ethics, Fairness, Inclusivity, and Code of Conduct.

Code: https://github.com/microsoft/bayesianize

Community Implementations: [ 1 code implementation](https://www.catalyzex.com/paper/arxiv:2105.14594/code)